Science convergence in affective research is associated with impactful multidisciplinary appeal rather than multidisciplinary content

The paper examines how convergence manifests and generates impact in affective (affectivism) versus cognitive (cognitivism) research. Historically, behaviorism dominated early 20th-century psychology, minimizing roles for cognition and affect. Cognitivism emerged mid-century, often linked to brain science and viewed as convergence-oriented, yet typically neglecting affective processes. Since the late 20th century, affective sciences have risen to address emotions and affect, with the affectivism framework proposing that affective processes are central to explaining cognition and behavior. A recent consensus suggested affectivism is growing and increasingly multidisciplinary. The study frames a core research question: What is the impetus of convergence science—does convergence feed upon convergence? The authors differentiate publication content categories—Affective, Cognitive, and Mixed (drawing from both)—and distinguish convergent content within a paper from convergent interest generated downstream via citations, to clarify how convergence relates to scholarly impact across these neighboring fields.

The authors situate their study within the evolution from behaviorism to cognitivism and the rise of affective sciences/affectivism. They reference the view of cognitivism as linked to convergence (brain science as a melting pot) but note mixed results of convergence efforts in brain science, motivating examination in other domains. A consensus paper posits affectivism’s ascent and growing multidisciplinary interactions. Prior science-of-science work highlights the role of topical composition and collaboration in citation impact and proposes best practices for citation normalization. The study builds on controlled vocabularies (MeSH) and diversity indices from interdisciplinarity research to operationalize thematic diversity and convergence, integrating article-level factors known to influence citations.

Data source and scope: PubMed (queried May 2023; 34,944,599 records total) was used due to comprehensive coverage and authoritative MeSH annotation. Two psychology co-authors selected MeSH terms representing affective and cognitive topics. Publications (1950–2022) annotated with at least one major MeSH term from the affective set were labeled Affective (n=314,665); those with at least one from the cognitive set were labeled Cognitive (n=314,968); those with at least one from each set were labeled Mixed (n=19,400). Total n=649,033 publications; 6.15% (39,972) had zero references and/or citations and were retained. Thematic consolidation: Only major MeSH terms were used. The MeSH hierarchy’s 14 relevant branches were aggregated into five subject areas (SA): SA1 Biological Sciences [A,B,G]; SA2 Psychological Sciences [F]; SA3 Medical Sciences [C,N]; SA4 Technical Methods [D,E,J,L]; SA5 Humanities [H,I,K,M]. Multidisciplinarity metrics: For each publication p, construct a 5-element vector e(p) counting major MeSH terms in SA1–SA5. Compute the outer product e(p)e(p)ᵀ, retain the upper triangular matrix, and normalize by the sum of its elements to obtain D~(p). Define thematic diversity D_p(p) as the proportion of off-diagonal weight in D~(p), i.e., the probability two randomly selected terms fall in different SAs (multidisciplinary content). For each citation c of p, compute D(c); define multidisciplinary appeal D_c(p) as the mean D(c) over all citations of p. Citation data and normalization: Citation and reference links were obtained via NIH iCite (linked to PubMed). To correct for temporal bias, citations c(p) were log-transformed and standardized within publication year t: CN(p) = [ln(c(p)+1) − mean_t(ln c)] / sd_t(ln c), yielding approximately standard normal distributions per year. Models: (1) Multiple linear regression predicting normalized citation impact CN(p): predictors included publication type dummies T_A (Affective), T_C (Cognitive) with Mixed as reference; D_p(p); D_c(p); ln A(p) (log number of authors); and binary controls F_SA1…F_SA5 indicating presence of any major MeSH in each SA. (2) Multinomial logistic regression predicting publication type (Affective or Cognitive vs Mixed) from D_p(p) and D_c(p), to compare structural differences in content diversity and downstream appeal across groups. Descriptive summaries: A(p) is right-skewed; ln A mean 1.22 ± 0.67. D_p mean 0.39 ± 0.17; D_c mean 0.43 ± 0.21. Subject-area aggregation reduced skew in MeSH-term counts. Visual analyses contrasted SA occurrence profiles among groups.

• Impact ranking by publication type: Mixed publications have the highest normalized citation impact, Affective are next, and Cognitive are lowest (Fig. 2A). Relative to Mixed, Affective show ~−4.5% and Cognitive ~−6.5% nominal citation differences, holding other factors constant.
• Multidisciplinary content vs appeal: In the regression (Eq. 4), thematic diversity of the paper’s content D_p is negatively associated with impact (β_Dp = −0.292, SE=0.017, p<0.001), whereas the mean diversity of citing papers D_c is strongly positively associated (β_Dc = 2.518, SE=0.006, p<0.001). Thus, broad downstream appeal matters more for impact than broad within-paper content.
• Authorship: Collaboration correlates positively with impact (β_lnA = 0.223, SE=0.001, p<0.001); a 100% increase in authors corresponds to ~22.3% more citations.
• Subject areas: SA1 (Biological Sciences) presence is positively associated with impact (β=0.099, p<0.001). SA2 (Psychological Sciences) shows no significant effect. SA3 (Medical; β=−0.110), SA4 (Technical; β=−0.071), and SA5 (Humanities; β=−0.118) are negatively associated (all p<0.001). Given that the core of these literatures lies in SA1+SA2, added diversity from SA3–SA5 aligns with the negative D_p effect.
• Descriptive diversity: D_p mean 0.39 ± 0.17; D_c mean 0.43 ± 0.21, indicating citing papers tend to be more thematically diverse than the cited papers.
• Publication type structure (Eq. 5): Higher D_p increases odds of being Affective or Cognitive relative to Mixed (Affective β=1.577; Cognitive β=1.621; both p<0.001). Higher D_c decreases odds of being Affective or Cognitive relative to Mixed (Affective β=−0.945; Cognitive β=−1.264; both p<0.001). Thus, Mixed papers exhibit the greatest multidisciplinary appeal, followed by Affective, then Cognitive.
• Exemplars: Mixed and Affective exemplar papers show relatively focused content (often SA1+SA2) but are cited across diverse SAs (including SA3–SA5), whereas Cognitive exemplars show citations with similar diversity to the original paper.

The study directly addresses whether convergence is driven by convergent content or by downstream convergent uptake. Results indicate that impactful work in affectivism (especially Mixed, but also Affective) does not rely on high within-paper topical diversity. Instead, papers with focused psycho-biological content that yield broadly useful concepts and datasets achieve wide cross-field citation, driving impact. Conversely, increasing within-paper thematic diversity correlates with lower impact, likely due to dilution or challenges in audience targeting, whereas broad multidisciplinary appeal among citers strongly boosts impact. The Mixed group’s synthesis of SA1 and SA2 appears to maximize cross-field utility and visibility, while purely Cognitive work tends to attract more specialized, like-for-like citations within brain science communities, limiting broader reach. These findings suggest that convergence in affective research is more about generating widely transferable value that attracts diverse downstream applications than about embedding multiple disciplines within each paper’s content.

This work provides a large-scale, publication-level analysis of convergence in affective versus cognitive research. It shows that: (1) Mixed publications are most impactful, followed by Affective, then Cognitive; (2) greater within-paper thematic diversity is associated with lower impact, whereas greater diversity among citing papers is strongly associated with higher impact; and (3) affectivism’s rise is linked to its broader multidisciplinary appeal rather than multidisciplinary content. These results refine understanding of convergence mechanisms, suggesting that focused contributions with broad utility can seed downstream convergence and impact. Future research should investigate causal mechanisms (e.g., prospective designs or natural experiments), examine generalizability across additional domains and databases, and explore how research design and dissemination strategies can optimize downstream multidisciplinary uptake.

The analysis is observational and regression-based; causal claims cannot be made. The findings challenge the common assumption that convergent content is necessary to breed convergence, suggesting instead that tools and concepts—even from mono- or narrowly disciplinary work—can diffuse widely. The study relies on PubMed coverage and MeSH major-topic annotations and on aggregation to five subject areas, which, while reducing noise, may omit nuances of subfield distances. Journal-level metrics were not included due to ecosystem changes; although topical controls partially proxy venue orientation, residual confounding may remain.